ai-system-design
behavioral
interviews
system-design
Mock prompt bank
How to use
105 prompts: 40 system design , 25 AI system design , 15 low-level design , 25 behavioral/Staff . Each is tagged with difficulty and links to prep pages where they exist.
Difficulty: M = Senior-bar standard · H = Senior-hard / Staff standard · XH = Staff/Principal (ambiguous, multi-team, migration or deep distributed theory).
Staff twist: for any prompt, add one of: "How would you roll this out across 100 teams?" , "Cut the cost by 50%" , "Now make it multi-region active-active" , "You have 2 engineers and 3 months — what do you build first?"
Use them with a peer, a paid mock, or the AI mock interviewer ; score with the rubric .
System design (40)
Classic web-scale
#
Prompt
Diff
Tags
Prep
SD1
URL shortener with custom aliases, expiry and near-real-time click analytics
M
hashing, KV, analytics
case study
SD2
Pastebin with 10 GB/day uploads and expiring pastes
M
blob storage, TTL
storage and CDN
SD3
News feed / timeline for 300M DAU
H
fan-out, ranking, caching
case study
SD4
Chat system (1:1 + groups up to 500, presence, read receipts)
H
websockets, ordering
case study
SD5
Notification system (push/email/SMS, 1B/day, preferences)
M
queues, retries, dedupe
case study
SD6
Typeahead / search autocomplete for 50k QPS
M
tries, caching
search systems
SD7
Web crawler for 1B pages/month with politeness
H
frontier, dedupe, scheduling
data pipelines
SD8
Video upload and streaming platform (YouTube-lite)
H
transcoding, CDN, adaptive bitrate
storage and CDN
SD9
Photo sharing with feed and likes (Instagram-lite)
M
blob, feed, counters
caching
SD10
Google Docs-style collaborative editor
XH
OT/CRDT, presence, consistency
consistency models
#
Prompt
Diff
Tags
Prep
SD11
Distributed rate limiter across 3 regions for 2,000 services
H
token bucket, consistency
rate limiting
SD12
Distributed key-value store with tunable consistency
XH
replication, quorums, gossip
case study
SD13
Distributed job scheduler (cron at scale, exactly-once execution semantics)
H
leases, idempotency
case study
SD14
Metrics and monitoring system (1M series/s ingest, 13-month retention)
H
TSDB, downsampling
case study
SD15
Distributed message queue / Kafka-lite
XH
log, partitions, ISR
messaging and streaming
SD16
Distributed cache (Redis-cluster-like) with hot-key handling
H
consistent hashing, eviction
partitioning
SD17
Centralised logging platform for 5 TB/day
H
ingestion, indexing, tiering
observability and SLOs
SD18
Feature flag / config service with < 1 s propagation
M
push vs pull, consistency
reliability patterns
SD19
API gateway for a microservices estate (auth, routing, quotas)
H
proxies, authN/Z
load balancing
SD20
Distributed lock service (Chubby/ZooKeeper-like)
XH
consensus, fencing tokens
consensus and Raft
SD21
Blob storage service (S3-lite) with erasure coding
XH
durability, metadata
storage and CDN
SD22
Unique ID generator (Snowflake-like) for 1M IDs/s
M
clocks, coordination-free
scalability fundamentals
SD23
Multi-region active-active user-profile service with DR (RPO 0, RTO 5 min)
XH
replication, failover
multi-region and DR
SD24
Change-data-capture pipeline from 200 Postgres DBs to a lakehouse
H
CDC, schemas, ordering
data pipelines
Transactions, money and marketplaces
#
Prompt
Diff
Tags
Prep
SD25
Payment system for a marketplace with multiple PSPs, ledger, reconciliation
XH
idempotency, double-entry
case study
SD26
Ride hailing / proximity matching (Uber-lite)
H
geo-index, matching
case study
SD27
Ticket booking (concerts) with 1M users at on-sale
H
contention, queues, holds
consistency models
SD28
Hotel/flight reservation system with overbooking rules
H
inventory, sagas
sagas and outbox
SD29
E-commerce order management (cart → order → fulfilment)
M
sagas, events
messaging and streaming
SD30
Digital wallet with P2P transfers and fraud checks
XH
ledger, consistency, risk
payment system
SD31
Stock exchange order matching engine (low latency)
XH
single-writer, sequencing
reliability patterns
SD32
Ad click aggregation with exactly-once counts for billing
H
stream processing, dedupe
data pipelines
Domain / enterprise (logistics flavoured)
#
Prompt
Diff
Tags
Prep
SD33
Real-time container/shipment tracking for 10M active shipments with ETA updates
H
events, geo, fan-out
messaging and streaming
SD34
Global booking platform for a shipping line (capacity, pricing, allocations)
XH
inventory, consistency, multi-region
data architecture
SD35
Event-driven integration hub between 300 partner systems (EDI, APIs, files)
H
integration patterns, schemas
integration patterns
SD36
Multi-tenant SaaS platform with tenant isolation and noisy-neighbour control
H
tenancy, quotas
security and multi-tenancy
SD37
Document management with search, versioning and retention for regulated data
M
storage, search, compliance
search systems
Staff / Principal (ambiguous, organisational)
#
Prompt
Diff
Tags
Prep
SD38
Migrate a 15-year-old monolith (order management) to services without downtime; 12 teams involved
XH
strangler fig, data migration
legacy modernization
SD39
Company has 3 incompatible event platforms after acquisitions; design the convergence
XH
platform strategy, migration
team topologies
SD40
Define API standards and versioning for 400 internal APIs; how do you enforce without blocking teams?
XH
governance, contracts
API contracts and versioning
AI system design (25)
RAG and search
#
Prompt
Diff
Tags
Prep
AI1
Enterprise RAG assistant over 2M documents with document-level ACLs and citations
H
hybrid retrieval, ACL, evals
rag system
AI2
Semantic product search for an e-commerce catalogue of 50M items
H
embeddings, hybrid, ranking
AI search
AI3
Customer-support RAG bot that must never give wrong policy answers
M
grounding, refusal, HITL
rag system
AI4
Multi-tenant RAG platform offered to 500 B2B customers
XH
isolation, cost, noisy neighbour
advanced RAG
AI5
Q&A over contracts requiring multi-hop reasoning ("which customers are affected if port X closes?")
XH
GraphRAG, structured + unstructured
advanced RAG
Assistants and agents
#
Prompt
Diff
Tags
Prep
AI6
ChatGPT-style assistant for 10M users: conversations, memory, file uploads
H
streaming, memory, safety
chat assistant
AI7
Multi-agent platform for 50 internal teams calling internal APIs
XH
tool registry, identity, HITL
agent platform
AI8
Operations copilot that triages exceptions and takes approved actions (capstone)
XH
orchestration, evals, guardrails
capstone
AI9
Coding agent / automated code-review bot for 2,000 repositories
H
sandboxing, context, evals
coding agent
AI10
Email/inbox agent that drafts and sends replies on a user's behalf
H
lethal trifecta, approvals
guardrails and security
AI11
Voice agent for a call centre (sub-second turn latency)
XH
streaming ASR/TTS, latency budget
cost and latency
AI12
Deep-research agent that produces cited reports in 10 minutes
H
planning, parallel tools, verification
multi-agent systems
AI13
Agent memory service shared across products (preferences, facts, episodes)
H
memory, privacy, provenance
memory systems
AI14
Long-running procurement agent that waits days for approvals and supplier replies
H
durable execution, HITL
durable execution
#
Prompt
Diff
Tags
Prep
AI15
Multi-tenant LLM gateway for 40 teams and 5 providers with budgets and chargeback
H
routing, quotas, observability
LLM gateway
AI16
LLM evaluation and observability platform for the whole company
XH
offline/online evals, judges, tracing
evaluation platform
AI17
Prompt management and experimentation platform (versioning, A/B, rollback)
M
config, experimentation
LLM observability
AI18
Guardrails service used by every AI product (input/output/tool policies)
H
classifiers, latency, policy
guardrails and security
AI19
MCP server registry and gateway for an enterprise (discovery, auth, audit)
XH
MCP, OAuth, governance
MCP
Serving, documents, cost
#
Prompt
Diff
Tags
Prep
AI20
LLM inference platform for 3 open-weight models at 5k req/min
XH
GPUs, batching, autoscaling
LLM serving platform
AI21
Intelligent document processing for 1M trade documents/month
H
OCR, extraction, HITL review
document processing
AI22
Cut LLM spend of an existing product by 60% without quality loss
H
routing, caching, distillation
capacity and cost planning
AI23
Fine-tuning platform for internal teams (data, training, eval, deploy)
XH
LoRA, eval gates, registry
fine-tuning
AI24
Recommendation system for a marketplace home feed (classic ML + LLM features)
H
two-tower, ranking, features
recsys basics
AI25
Your company wants "AI in every product" in 12 months: design the shared platform and operating model
XH
strategy, platform, governance
AI-native architecture
Low-level design (15)
#
Prompt
Diff
Focus
Prep
LLD1
Parking lot (spots by size, tickets, pricing) + change request: EV charging with reservations
M
modelling, strategy pattern
design patterns
LLD2
LRU cache with TTL, thread-safe
M
data structures, locking
concurrency and LLD
LLD3
Rate limiter library (token bucket, sliding window) pluggable per key
M
strategy, concurrency
rate limiting
LLD4
Elevator control system for a 40-floor building with 6 elevators
H
state machines, scheduling
design patterns
LLD5
In-memory pub/sub message broker with topics, consumer groups, at-least-once
H
concurrency, queues
concurrency and LLD
LLD6
Splitwise-style expense sharing with debt simplification
M
modelling, graph
SOLID and refactoring
LLD7
Library management / book lending system
M
modelling, invariants
DDD tactical
LLD8
Shipment lifecycle state machine with guards, side effects and audit
H
state pattern, events
DDD tactical
LLD9
Thread pool / bounded executor with graceful shutdown
H
concurrency primitives
concurrency models
LLD10
Job scheduler (cron expressions, retries, priorities) in-process
H
heaps, time, concurrency
heap
LLD11
Logging framework with levels, appenders, async flushing
M
chain of responsibility, observer
design patterns
LLD12
Chess or tic-tac-toe game engine with undo
M
command pattern, validation
design patterns
LLD13
Tool-calling agent runtime: tool registry, schema validation, retries, budgets
H
interfaces, typing
tool calling
LLD14
Circuit breaker + retry with backoff library
H
state machine, concurrency
reliability patterns
LLD15
Vending machine / payment kiosk with multiple payment methods
M
state pattern
design patterns
Behavioral / Staff (25)
Prep: behavioral interviews . Answer in STAR(L): Situation (< 45 s), Task, Actions (with "I"), Result (numbers), Learning.
Scope and impact
#
Prompt
Diff
Signal tested
Related
B1
Tell me about your highest-impact work in the last 3 years.
H
scope, impact
Staff archetypes
B2
Describe the most technically complex project you led end-to-end.
M
technical depth, ownership
B3
Tell me about a technical strategy or vision you created for multiple teams.
XH
strategy, alignment
technical strategy
B4
Tell me about a time you identified a problem nobody had asked you to solve.
H
ownership, ambiguity
B5
Describe a time you reduced cost, latency or incidents significantly. How did you measure it?
M
impact, data
Influence and conflict
#
Prompt
Diff
Signal tested
Related
B6
Tell me about a conflict with a peer or another team over a technical decision.
M
conflict, collaboration
influence without authority
B7
Tell me about a time you influenced a decision outside your team without authority.
H
influence
influence without authority
B8
Tell me about a time you disagreed with your manager or a senior leader.
H
courage, judgement
communication
B9
Describe a time you had to get several teams to adopt a standard or platform.
XH
coalition-building
architecture reviews
B10
Tell me about a decision you made that was unpopular.
H
conviction, empathy
decision making
Ambiguity, judgement and failure
#
Prompt
Diff
Signal tested
Related
B11
Tell me about a project that failed or missed its goals.
M
self-awareness
B12
Tell me about a time you were wrong about a technical decision.
H
humility, learning
B13
Tell me about a time you made a decision with incomplete data.
H
judgement, reversibility
decision making
B14
Tell me about a time you killed or significantly changed a project.
XH
prioritisation
B15
How did you handle significant technical debt competing with feature work?
H
trade-offs, business framing
technical debt
B16
Describe a time you balanced speed against quality under a hard deadline.
M
pragmatism
Growing others and org health
#
Prompt
Diff
Signal tested
Related
B17
Tell me about someone you mentored or sponsored into a bigger role.
H
growing others
mentoring and sponsorship
B18
How have you raised the engineering bar beyond your own team?
XH
mechanisms
architecture reviews
B19
Tell me about a time you gave difficult feedback to a senior engineer.
H
candour
B20
Describe how you onboarded into a new, complex domain quickly.
M
learning
Operations, AI era and motivation
#
Prompt
Diff
Signal tested
Related
B21
Describe an incident you led. What changed afterwards?
H
incident leadership
incident leadership
B22
How have you introduced AI tools or AI features into a team responsibly? What went wrong?
H
AI-era leadership
AI-era leadership
B23
Tell me about a design doc or RFC you wrote that changed direction after review.
M
written communication
design docs and RFCs
B24
What kind of Staff engineer are you (archetype), and what work do you avoid?
H
self-knowledge
Staff archetypes
B25
Why this role, and what would you do in your first 90 days?
M
motivation, planning
Story bank coverage
Map your 8–10 strongest stories against B1–B25. Each story should cover 3+ prompts. Any prompt with no story → write one before the next checkpoint.